| image-classification-on-omnibenchmark | SwinTransformer | #2 | Average Top-1 Accuracy: 46.4 |
| instance-segmentation-on-occluded-coco | Swin-B + Cascade Mask R-CNN | #2 | Mean Recall: 62.90 |
| instance-segmentation-on-occluded-coco | Swin-S + Mask R-CNN | #5 | Mean Recall: 61.14 |
| instance-segmentation-on-occluded-coco | Swin-T + Mask R-CNN | #6 | Mean Recall: 58.81 |
| instance-segmentation-on-separated-coco | Swin-B + Cascade Mask R-CNN | #2 | Mean Recall: 36.31 |
| instance-segmentation-on-separated-coco | Swin-S + Mask R-CNN | #5 | Mean Recall: 33.67 |
| instance-segmentation-on-separated-coco | Swin-T + Mask R-CNN | #6 | Mean Recall: 31.94 |
| semantic-segmentation-on-foodseg103 | Swin-Transformer (Swin-Small) | #4 | mIoU: 41.6 |
| instance-segmentation-on-coco-test-dev | Swin-L (HTC++, multi scale) | #19 | mask AP: 51.1 |
| instance-segmentation-on-coco-test-dev | Swin-L (HTC++, single scale) | #21 | mask AP: 50.2 |
| instance-segmentation-on-coco-minival | Swin-L (HTC++, multi scale) | #24 | mask AP: 50.4 |
| instance-segmentation-on-coco-minival | Swin-L (HTC++, single scale) | #29 | mask AP: 49.5 |
| object-detection-on-coco-test-dev | Swin-L (HTC++, multi scale) | #31 | box mAP: 58.7 |
| object-detection-on-coco-test-dev | Swin-L (HTC++, single scale) | #33 | box mAP: 57.7 |
| thermal-image-segmentation-on-mfn-dataset | SwinT | #34 | mIOU: 49.0 |
| semantic-segmentation-on-ade20k-val | Swin-L (UperNet, ImageNet-22k pretrain) | #38 | mIoU: 53.5 |
| semantic-segmentation-on-ade20k-val | Swin-B (UperNet, ImageNet-1k pretrain) | #53 | mIoU: 49.7 |
| object-detection-on-coco-val2017 | Swin-L (HTC++, multi scale) | #60 | box AP: 58 |
| object-detection-on-coco-val2017 | Swin-L (HTC++, single scale) | #73 | box AP: 57.1 |
| semantic-segmentation-on-ade20k | Swin-L (UperNet, ImageNet-22k pretrain) | #83 | Validation mIoU: 53.50Test Score: 62.8 |
| semantic-segmentation-on-ade20k | Swin-B (UperNet, ImageNet-1k pretrain) | #136 | Validation mIoU: 49.7 |
| image-classification-on-imagenet | Swin-L | #103 | Top 1 Accuracy: 87.3%Number of params: 197MGFLOPs: 103.9 |
| image-classification-on-imagenet | Swin-B | #157 | Top 1 Accuracy: 86.4%Number of params: 88MGFLOPs: 47 |
| image-classification-on-imagenet | Swin-T | #688 | Top 1 Accuracy: 81.3%Number of params: 29MGFLOPs: 4.5 |